Molten salts chemistry from lab to applications

pages cm Summary: “In recent years, molecular modelling has become an indispensable tool for studying the structure and dynamics of molten salts.. Benoit Claux European Commission, Joint

Molten Salts Chemistry From Lab to Applications

Intentionally left as blank

Molten Salts Chemistry From Lab to Applications

30 Corporate Drive, Suite 400, Burlington, MA 01803, USA

525 B Street, Suite 1800, San Diego, CA 92101-4495, USA

First edition 2013

©2013 Elsevier Inc All rights reserved.

No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or

by any means electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher.

Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone ( þ44) (0) 1865 843830; fax (þ44) (0) 1865 853333; email: permissions@elsevier.com Alternatively you can submit your request online by visiting the Elsevier web site at http://elsevier.com/locate/permissions , and selecting Obtaining permission to use Elsevier material

Notice

No responsibility is assumed by the publisher for any injury and/or damage to persons or property as

a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein Because of rapid advances in the medical sciences,

in particular, independent verification of diagnoses and drug dosages should be made.

Library of Congress Cataloging-in-Publication Data

Molten salts chemistry: from lab to applications / edited by Fre´de´ric Lantelme, Henri Groult – First edition pages cm

Summary: “In recent years, molecular modelling has become an indispensable tool for studying the structure and dynamics of molten salts In this chapter we first provide a short description of the state-of-the-art models and methods used for modelling molten salts at the atomic scale In particular, we discuss the importance of polarization effects for obtaining accurate results We then give some examples of the structure of several molten salts, as yielded by the simulations We finish by describing how the transport properties, which encompass the diffusion coefficients, electrical conductivities, viscosities and thermal conductivities, are computed By comparing the values given by the simulations to reference experimental data, we show that this technique can now be considered as highly predictive”– Provided by publisher Includes bibliographical references and index.

A catalogue record for this book is available from the British Library

For information on all Elsevier publications

visit our web site at store.elsevier.com

Printed and bound in USA

13 14 15 16 17 10 9 8 7 6 5 4 3 2 1

ISBN: 978-0-12-398538-5

M Salanne, C Simon, P Turq, N Ohtori, and P.A Madden

2 Raman Spectroscopy and Pulsed Neutron Diffraction of Molten

Salt Mixtures Containing Rare-Earth Trichlorides: Trial Approaches

Catherine Bessada and Anne-Laure Rollet

3.2 Experimental Techniques: Specificity, Limitation, Setup 34

4 Thermodynamic Calculations of Molten-Salt Reactor Fuel Systems 49

O Benesˇ and R.J.M Konings

5 Ionic Transport in Molten Salts 79Isao Okada

Fre´de´ric Lantelme, Henri Groult, Hugo Mosqueda, Pierre-Louis Magdinier,Herve´ Chavanne, Vincent Monteux, and Philippe Maurin-Perrier

Soghomon Boghosian and Rasmus Fehrmann

7.2 Physicochemical Properties of the Catalyst Model System 1347.3 Phase Diagrams of Molten Binary Systems of Relevance to the

7.4 Multi-instrumental Investigations and Complex Formation in

7.5 Activity and Deactivation of SO2Oxidation Vanadia–Pyrosulfate

Bulk Melts and Supported Molten Salts: Formation of Crystalline

7.6 Vanadium Crystalline Compound Formation: A Summary of

Structural and Vibrational Properties and Implications of Catalytic

7.7 In Situ Spectroscopy of Catalyst Models and Industrial Catalysts 1477.8 Mechanism of the SO2Oxidation Catalytic Reaction 149

9.4 Carbon as an Inert Anode in the Absence of Oxygen

10.3 Thermodynamics of Oxygen Electrode Reaction on a

Akimasa Tasaka

11.2 Anodic Behavior of Nickel and Nickel-Based Composite

Electrodes in NH4F
2HF at 100C for Electrolytic Production

11.3 Anodic Behavior of Carbon Electrode in NH4F
KF
mHF

(m¼3 and 4) at 100C for Electrolytic Production of NF

11.4 New Development for Electrolytic Production of NF3

K Sridharan and T.R Allen

14.5 LiF(80.5)-CaF2(19.5) Melts 298

15 Electrochemical Synthesis of Novel Niobium and Tantalum

T Brousse, P Simon, B Daffos, S Komaba, and N Kumagai

16.1 Synthesis of Carbon Nanopowders (CNPs) in Molten

18 Synthesis and LiþIon Exchange in Molten Salts of Novel

Hollandite-Type Ky(Mn1xCox)O2
zH2o Nanofiber for Lithium Battery

20.2 Physicochemical Properties and Corrosion Aspects of Molten

20.3 Molten Salt Thermal Energy Storage Applications for Concentrated

21 The Sodium Metal Halide (ZEBRA) Battery: An Example of

Akane Hartenbach, Michael Bayer, and Cord-Henrich Dustmann

21.2 Battery-Relevant Properties of the Molten Salt Electrolyte 43921.3 Involvement of the Molten Electrolyte in Battery’s Safety

21.4 Future Use of the ZEBRA Technology in Grid Applications 448

22 Hydrogen Storage and Transportation System through

Yuzuru Sato and Osamu Takeda

23.4 Accelerator Molten Salt Breeder for233U production 48923.5 Regional Center for Chemical Processing and Fissile Production 491

24.3 Processes in Progress for Future Nuclear Applications

26 Development of Pyrochemical Separation Processes for

Recovery of Actinides from Spent Nuclear Fuel in Molten LiCl-KCl 541Jean-Paul Glatz, Rikard Malmbeck, Pavel Soucˇek, Benoˆıt Claux,

Roland Meier, Michel Ougier, and Tsuyoshi Murakami

Michael Bayer Battery Consult GmbH, Zeughaustrasse 19D, Meiringen, Switzerland.

O Benesˇ European Commission, Joint Research Centre, Institute for TransuraniumElements, P.O Box 2340, Karlsruhe, Germany

Catherine Bessada CNRS, Universite´ d’Orle´ans, UPR 3079, Laboratoire CEMHTI,

Andrew C Chien School of Chemistry, University of St Andrews, St Andrews, Fife,United Kingdom

Benoit Claux European Commission, Joint Research Centre, Institute forTransuranium Elements, 76125 Karlsruhe, Germany

B Daffos Universite´ Paul Sabatier, CIRIMAT, UMR CNRS 5085, 31062 Toulousecedex 4, France.

Sylvie Delpech IPNO, Universite´ Paris Sud, 91406 Orsay Cedex, Paris, France.Cord-Henrich Dustmann Battery Consult GmbH, Zeughaustrasse 19D, Meiringen,Switzerland

Markus Eck German Aerospace Center (DLR), Institute of TechnicalThermodynamics, Stuttgart, Germany

Rasmus Fehrmann Department of Chemistry, the Technical University of Denmark,Lyngby, Denmark

Mathieu Gibilaro Laboratoire de Ge´nie Chimique (LGC), De´partementProce´de´s Electrochimiques UMR 5503, Universite´ Paul Sabatier, Toulouse cedex 9,France

Jean-Paul Glatz European Commission, Joint Research Centre, Institute forTransuranium Elements, 76125 Karlsruhe, Germany

Takuya Goto Department of Environmental Systems Science, Doshisha University,Kyotanabe, 610-0321, Kyoto, Japan

Henri Groult Laboratoire PECSA, CNRSUMR 7195, Universite´ Pierre et Marie Curie,

4 place Jussieu, 75252, Paris Cedex 05 France

Rika Hagiwara Department of Fundamental Energy Science, Kyoto University, Sakyo,606-8056 Kyoto, Japan

Akane Hartenbach Battery Consult GmbH, Zeughaustrasse 19D, Meiringen,Switzerland

John T.S Irvine School of Chemistry, University of St Andrews, St Andrews, Fife,United Kingdom

Yasuhiko Ito Energy Conversion Research Center, Doshisha University, Kyotanabe,Kyoto, Japan

Yasuhiko Iwadate Graduate School of Engineering, Chiba University, Chiba,Japan

C.M Julien UPMC Universite´ Paris 06, UMR 7195, Laboratoire PECSA, F-75005,Paris, France

Yuya Kado Department of Fundamental Energy Science, Kyoto University, Sakyo,606-8056 Kyoto, Japan.

Y Kadoma Iwate University, Morioka, Iwate, Japan

Stefanie Kaesche Materials Testing Institute University of Stuttgart (MPA), Stuttgart,Germany

M Kawase Energy Engineering Research Laboratory, Central Research Institute ofElectric Power Industry, Yokosuka-Shi, Japan

S Komaba Department of Applied Chemistry, Tokyo University of Science,Kagurazaka 1-3, Shinjuku, Tokyo 162-8601, Japan

R.J.M Konings European Commission, Joint Research Centre, Institute forTransuranium Elements, P.O Box 2340, Karlsruhe, Germany

N Kumagai Faculty of Engineering, Iwate University, Morioka, Japan

S.A Kuznetsov I.V Tananaev Institute of Chemistry and Technology of Rare Elementsand Mineral Raw Materials, Kola Science Centre of the Russian Academy of Sciences,Murmansk Region, Russia

Doerte Laing German Aerospace Center (DLR), Institute of TechnicalThermodynamics, Stuttgart, Germany

Virginie Lair Chimie ParisTech ENSCP, UMR CNRS 7575, Laboratory ofElectrochemistry, Chemistry of Interfaces and Modelling for Energy, Paris, France.Fre´de´ric Lantelme Laboratoire PECSA, CNRS UMR 7195, Universite´ Pierre et MarieCurie, 4 place Jussieu, 75252, Paris Cedex 05 France

K Le Van UPMC Universite´ Paris 06, UMR 7195, Laboratoire PECSA, F-75005,Paris, France

P.A Madden Department of Materials, University of Oxford, Oxford, United Kingdom.Pierre-Louis Magdinier Institut de Recherches En Inge´nierie des Surfaces, ZI Sud-rueBenoıˆt Fourneyron, CS42077, 42162, Andre´zieux Bouthe´on, France

Rikard Malmbeck European Commission, Joint Research Centre, Institute forTransuranium Elements, 76125 Karlsruhe, Germany

Laurent Massot Laboratoire de Ge´nie Chimique (LGC), De´partement Proce´de´sElectrochimiques UMR 5503, Universite´ Paul Sabatier, Toulouse cedex 9, France

Philippe Maurin-Perrier Institut de Recherches En Inge´nierie des Surfaces, ZISud-rue Benoıˆt Fourneyron, CS42077, 42162, Andre´zieux Bouthe´on, France.Roland Meier European Commission, Joint Research Centre, Institute forTransuranium Elements, 76125 Karlsruhe, Germany.

Vincent Monteux Institut de Recherches En Inge´nierie des Surfaces, ZI Sud-rue BenoıˆtFourneyron, CS42077, 42162, Andre´zieux Bouthe´on, France

Hugo Mosqueda Laboratoire PECSA, CNRS UMR 7195, Universite´ Pierre et MarieCurie, 4 place Jussieu, 75252, Paris Cedex 05 France

Tsuyoshi Murakami Central Research Institute of Electric Power Industry, 2-11-1Iwado-kita, Komae-shi, Tokyo 201-8511, Japan

Tokujiro Nishikiori R&D Division, I’ MSEP Co., Ltd., Kyotanabe, Kyoto, Japan.Toshiyuki Nohira Department of Fundamental Energy Science, Graduate School ofEnergy Science, Kyoto University, Kyoto, Japan

N Ohtori Graduate School of Science and Technology, Niigata University, Niigata,Japan

Isao Okada Tokyo Institute of Technology, O-okayama 2-12-1, Meguro-ku, Tokyo152-8550, Japan

Michel Ougier European Commission, Joint Research Centre, Institute forTransuranium Elements, 76125 Karlsruhe, Germany

Nicole Pfleger German Aerospace Center (DLR), Institute of TechnicalThermodynamics, Stuttgart, Germany

Armelle Ringuede´ Chimie ParisTech ENSCP, UMR CNRS 7575, Laboratory ofElectrochemistry, Chemistry of Interfaces and Modelling for Energy, Paris, France.Anne-Laure Rollet CNRS, UPMC Universite´ Paris 06, UMR 7195, LaboratoirePECSA, F-75005 Paris, France

M Salanne UPMC Universite´ Paris 06, CNRS, ESPCI, UMR 7195, LaboratoirePECSA, Paris, France

Yuzuru Sato Department of Metallurgy, Tokoku University, Sendai 980–8579, Japan

C Simon UPMC Universite´ Paris 06, CNRS, ESPCI, UMR 7195, Laboratoire PECSA,Paris, France

P Simon Universite´ Paul Sabatier, CIRIMAT, UMR CNRS 5085, 31062 Toulousecedex 4, France.

Pavel Soucˇek European Commission, Joint Research Centre, Institute forTransuranium Elements, 76125 Karlsruhe, Germany

K Sridharan Department of Engineering Physics, University of Wisconsin–Madison,Madison, Wisconsin

Wolf-Dieter Steinmann German Aerospace Center (DLR), Institute of TechnicalThermodynamics, Stuttgart, Germany

Osamu Takeda Department of Metallurgy, Tokoku University, Sendai 980–8579,Japan

Akimasa Tasaka Department of Molecular Chemistry and Biochemistry, Faculty ofScience and Engineering, Doshisha University,1-3 Miyako-dani, Tadara, Kyotanabe,Kyoto 610-0321, Japan

Pierre Taxil Laboratoire de Ge´nie Chimique (LGC), De´partement Proce´de´sElectrochimiques UMR 5503, Universite´ Paul Sabatier, Toulouse cedex 9, France.Manabu Tokushige Department of Materials Science and Engineering, NorwegianUniversity of Science and Technology, Trondheim, Norway

P Turq UPMC Universite´ Paris 06, CNRS, ESPCI, UMR 7195, Laboratoire PECSA,Paris, France

Dihua Wang School of Resource and Environmental Science, Wuhan University,Wuhan 430072, China

Wei Xiao School of Resource and Environmental Science, Wuhan University, Wuhan

430072, China

Ritsuo Yoshioka International Thorium Molten-Salt Forum (ITMSF) 3-17-24,Hino-chuou, Konan-ku, Yokohama, 234-0053, Japan

Intentionally left as blank

Fused salts are widely used in many industrial processes needing to free the limitations ing from the use of aqueous solutions In connection with their exceptional properties, fusedmedia offer a vast panel of uses: their wide potential window between decomposition limitsallows the electro-winning of highly electropositive elements or the preparation of very elec-tronegative elements Their thermal stability and generally low vapor pressure are welladapted to high-temperature chemistry, enabling fast reaction rates Their ability to dissolvemany inorganic compounds such as oxides, nitrides, carbides, and other salts makes themideal solvents useful in electrometallurgy, metal coating, treatment of by-products, andenergy conversion It is recalled that one of the most important chemicals produced inthe world, sulfuric acid, is made by a molten salt catalysis

aris-This monograph summarizes the recent advances on these topics, maintaining a linkbetween fundamental investigations and industrial developments However, to limit the booksize, the well-known classical applications of fused salts in the aluminum and fluorine prep-aration are not described here; many well-documented books on these subjects are available

in the recent literature The aim of the book is to present the state of the art of the currentresearches performed by the molten salt community, in the spirit of the invaluable organi-zation work carried out by our colleague Marcelle Gaune-Escard

Molten salts play a major role in the development of energy resources Since many years,the reprocessing of nuclear wastes has become a priority for nations using nuclear energy; inthat domain, different pyrochemical devices have been investigated involving molten saltsolvents Moreover, they appear as a promising route toward the emergence of a safer nuclearenergy (nuclear reactors, Generation IV)

Now, laboratory research using fused salts opens ways for interesting applications Theycan be used in some new processes such as the NF3production Plasma-induced dischargeelectrolysis in fused electrolytes is an attractive novel approach to producing functionalnanoparticles Molten carbonate/solid oxide fuel cells appear as a promising technology

to realize direct conversion of solid carbon to electricity Materials for energy storagedevices can be successfully prepared by fused salt electrolysis: carbon nanoparticles withvery large specific surface for efficient supercapacitors are obtained in fused alkali carbon-ates New cathode materials for rechargeable lithium batteries are generated from synthesisand ion exchange in molten LiNO3-LiCl of hollandite-type ∝-MnO2 High-temperaturemolten salt batteries are also studied for high-capacity energy storage Fused alkalinitrates/nitrites are valuable materials for heat transport and storage in solar plants.Molten salt bathes remain always of large use in industry They are recognized as superiorprocesses for heat-treating a variety of metals from austempered ductile iron to high-speedtool steel and also nonmetals, such as glass, plastics, and rubber They remain privilegedmedia for the surface treatments of tool steels including nitriding, nitrocarburizing, boriding,and other steel surface hardening methods Indeed, this technology offers invaluable advan-tages which are briefly described Fundamental reactions and chemical behavior of surfacelayers in fused salts are examined

These concrete applications have induced a renewed interest for a fundamental study ofthe specific features of high-temperature ionic liquids and thus some chapters devoted to thisdescription are included in the book.

The book contains 26 chapters written by authors all recognized as specialists activelyworking in fused salt chemistry, electrochemistry, and catalysis We hope it offers newaspects of molten salt chemistry to readers belonging to academic and industrial world Itshould be useful for generating new ideas showing the interest of the fused salt route

Fre´de´ric LantelmeHenri Groult (Eds.)

1 Modeling of Molten Salts

M Salanne * , C Simon * , P Turq * , N Ohtori{,

P.A Madden{

*UPMC Univ Paris 06, CNRS, ESPCI, UMR 7195, Laboratoire PECSA, Paris, France,

{Graduate School of Science and Technology, Niigata University, Niigata,Japan,{Department of Materials, University of Oxford, Oxford, United Kingdom

1.1 Introduction

Among the large array of techniques, which are devoted to the study of liquid matter,molecular simulations appear as a method of choice [1] They provide an atomic-scaledescription of the systems and are therefore used both for interpreting existing experimen-tal data and for predicting unknown properties In the case of molten salts, this ability

is crucial Indeed, due to the high temperatures and sometimes due to the use of hazardous

or radioactive species, experiments in these media are expensive and difficult to realize,

if not impossible Simulation is then an alternative method which allows a widerange of thermodynamic (e.g., temperature) and composition ranges to be spanned for

to their changing coordination environments, allowing for a compact representation ofmany-body contributions (e.g., polarization) to the interaction energy [3] Such potentialsare erroneously termed “empirical,” although it is only appropriate when experimental infor-mation is used in the parameterization procedure In molten salts, it is possible to obtain thepotential parameters by fitting the predicted forces and multipoles to a large body of infor-mation generated fromab initio calculations [4–6]

Of course it is not because a calculation is entirely based on first principles that it providesthe correct answer Any method involves some more-or-less controlled approximations [7],and real systems often involve small impurities, interfaces with vessel furnaces, etc., whichare not included in the simulations Testing the simulations on a set of reliable data is com-pulsory As soon as this step has successfully been taken, it is possible to use them in a pre-dictive way The properties of interest are usually the structure of the melts along with theirenthalpy of mixing, heat capacity, and density (i.e., “static” properties) or their diffusioncoefficients (one for each species), electrical conductivity, viscosity, and thermal conductiv-ity (“transport” or “dynamic” properties) The list of known quantities varies importantlyfrom one system to another, and molecular simulations are techniques of choice for (i) fillingMolten Salts Chemistry

© 2013 Elsevier Inc All rights reserved.

the gaps in the databases and (ii) interpreting the data and linking them one with each other(e.g., linking physical properties of the melt with its structure).

This chapter is organized as follows: In a first section, we will provide a brief summary ofthe principal methods for performing molecular simulation and of the models which arewidely used for simulating molten salts Then we will provide a state-of-the-art picture ofmolten salts in terms of structural, static, and dynamic properties In each of these sections,the basic methodological aspects for extracting the useful information will be explained and

a few selected examples will be provided In the conclusion, future directions for the ing of molten salts will be proposed

model-1.2 Methods and Models

1.2.1 Molecular Dynamics Simulations

Molecular dynamics is a widespread computational technique in which systems are lated at the atomic scale [1] In the case of liquids, a simulation cell typically contains100-10,000 atoms (even more for complex systems), and periodic boundary conditionsare used; an example is shown inFigure 1.1for molten LiF-KF (the lines correspond tothe simulation cell limits, at which the periodic boundary conditions are applied) A trajec-tory of several nanoseconds is then calculated by numerically solving Newton’s equation ofmotion

in the framework of a given model The models used in the case of molten salts are described

in the following It is worth noting that several statistical mechanics ensembles can beused for simulating a system In the simplest case, the number of atomsN, the volume ofthe simulation cellV, and the total energy of the system E are kept fixed (microcanonicalensemble or NVE ensemble) When the system is at the thermodynamic equilibrium, the

Figure 1.1 Typical simulation cell (equimolar

LiF-KF mixture) Ions are shown as spheres, while

the lines represent the limits of the cell, where

periodic boundary conditions are applied.

temperature T and pressure P then fluctuate around an average value It is neverthelessoften very useful to control the temperature by adding a thermostat; then the total energy

is no more constant and the system is simulated in the canonical ensemble (or NVTensemble) This is what has been done in most of the results presented here Finally,one can also control the pressure, in which case the volume is allowed to fluctuate Suchsimulations, performed in the NPT ensemble, are mainly used for determining the equa-tion of state and the constant pressure heat capacity of the system

1.2.2 The Rigid Ion Model

In a first approximation, ions in a molten salt can be represented as a set of charged sphereswith a rigid electron density [8,9] In such a rigid ion model (RIM), the interaction potential isthe sum of three terms:

The first one accounts for the Coulombic interactions:

The second term is due to the overlap repulsion between the electronic clouds, it reads

The last term, the dispersion, arises from correlated fluctuations of the electrons; it isalways attractive and it takes the form

where C6and C8are the dipole-dipole (dipole-quadrupole) dispersion coefficients andfn

(n¼6 or 8) are functions that can be introduced to describe the short-range penetrationcorrection to the asymptotic multipole expansion of dispersion In our work, we useTang-Toennies functions [11], which take the following form:

where thebnparameter sets the range of the damping effect The dispersion term accounts for

a small part only of the total force acting on an ion; nevertheless, it plays a crucial role in thedensity predicted by the model [12]

1.2.3 The Polarizable Ion Model

Although the RIM provides a correct description of the interaction for a few simple moltensalts [13–15], a vast majority of the useful systems involve multiply charged ions (mostlycations), for which it is not accurate enough A great enhancement of the reliability ofthe simulations can be obtained by including additional polarization effects in the model[3,16] This leads to the polarizable ion model (PIM), which contains the following term:

(1.6)

whereairepresents the (scalar) polarizability of ioni andmiits induced dipole Note that theformer are additional parameters, while the dipoles are treated as additional degrees of free-dom which are allowed to fluctuate during the simulations [17] They are given by solving, ateach time step, self-consistently the set equations

in whichEiis the electric field felt by ioni, which is due not only to the distribution of othercharges but also to the distribution of all the induced dipole moments Solving the set ofEquation(1.7)is formally equivalent to minimizing the polarization term, and very oftenthis is what is done in practice In Equation(1.6), we also introduce some Tang-Toenniesfunctions,

which account for the short-range penetration effects

The importance of polarization effects is schematized for BeF2inFigure 1.2 The twodoubly charged cations strongly repel each other In the case of the RIM, this repulsion favorsthe formation of a Be-F-Be arrangement where the fluoride ion lies on the Be-Be axis, inorder to maximize the screening between them In the PIM, an additional induced dipole

is created on the anion An additional flexibility is then gained: the Be-F-Be angle cannow depart from 180, with the dipole pointing toward its bisector The electronic cloud

of the anion is then shifted with respect to the position of the nuclei, which effectively screensthe cation-cation repulsion

In most of the systems of technological interest, multiply charged species are present[18–21]; therefore, in the following, we will only describe results obtained in the framework

of the PIM

1.2.4 Interaction Potential Parameterization

Molecular dynamics simulations can be predictive provided that no empirical information isused in their construction and highly transferable from pure materials to mixtures Our PIMpotentials are parameterized by a generalized “force-matching” method A suitablecondensed-phase ionic configuration is taken from a molecular dynamics simulation usingsome approximate force field for the material of interest Typically, 100 ions would be used

in periodic boundary conditions The configuration is then input to a plane-wave density

functional theory electronic structure program, and an energy minimization is carried out tofind the ground-state electronic structure [22] It is important that this electronic structurecalculation is made on a condensed-phase ionic configuration, as the electronic structure

of an anion is strongly affected by the confining potential produced by the proximity ofits coordinating cations [3,12] From the results of this calculation, the force and dipolemoment on each ion are obtained, the latter by making use of the transformation of theKohn-Sham orbitals to a maximally localized Wannier function set [23] The parameters

in the polarizable potential are then optimized by matching the dipoles and forces fromthe potential on the same ionic configuration to theab initio values [4,6] An example ofresult obtained from such a fitting procedure is shown inFigure 1.3

If necessary, the process may be iterated, by using the fitted potential to generate a newionic configuration to input to theab initio calculation The resulting potentials can then beused in much larger scale molecular dynamics simulations to obtain the physical properties

of interest In the case of simple systems, the “force matching” may even be avoided byputing the various interaction terms separately [5,6]

com-Figure 1.2 Importance of polarization effects.

In the RIM, the central fluoride ion stands between the two beryllium cations in order to screen their electrostatic repulsion In the PIM, the formation

of an induced dipole results in additional screening, when the fluoride ion lies off the line of centers of the berylliums, which favors the occurrence of bent Be-F-Be configurations.

Figure 1.3 The quality of the fit of the forces on the ions to those obtained from the ab initio calculation is shown for one configuration containing 100 ions The squares show the values obtained with the PIM potential (dashed line is a guide to the eye), and the circles results from the ab initio calculations.

1.2.5 Calculated Quantities

During a molecular simulation, the iterative integration of Newton’s equation of motion vides a trajectory of the particles, i.e., the evolution of their positionsriand velocitiesvi Thestructural arrangement is straightforwardly deduced from the instantaneous positions, but theextraction of the thermodynamic and dynamic properties requires to calculate many otherquantities Here is a (nonexhaustive) list of what can be calculated during a moleculardynamics simulation:

pro-– Kinetic energy

– Potential energy

– Temperature (in the NVE ensemble)

– Pressure (in the NVE and NVT ensembles)

– Volume (in the NPT ensembles)

– Stress tensor

– Energy current

1.3 Structure of Molten Salts

The structure of any molten salt is characterized by an alternation of positively and tively charged ionic solvation shells around a given ion [13] This arises from the predom-inance of Coulombic effects, which results in a strong attraction between oppositely chargedspecies and a strong repulsion otherwise At very short range, the overlap repulsion remains

nega-of course stronger: the shortest interionic distances (at “contact”) are thus due to this action The partial radial distribution functions, which describe how the density of a speciesvaries as a function of distance from another species, are characterized by strong oscillations,which persist up to distances that are (much) larger than in the case of other nonionic solventssuch as water [24] These functions are typical output of molecular dynamics simulations,and they are related to the diffraction patterns which are recorded in neutron and X-rayexperiments In the following, we will therefore base our discussion on the structure on suchfunctions

inter-Despite sharing such common features, the structure of molten salts can vary ably depending on the relative sizes and charges of the ions Due to the large variety of mol-ten salts that can be formed, it is very difficult to draw some general laws, but if we restrainthe analysis to so-called pure molten salts (i.e., which consist of two species, one cation andone anion only; for an extensive review on the structure of molten salts—including multi-component ones—the reader is referred to Ref [25]), it is possible to observe some particulartrends In order to underline these, we show inFigure 1.4the partial radial distribution func-tions for three pure molten salts, namely, LiF, BeF2, and AlF3 The LiF ones show the typicalshape for alkali halides: The anion-anion and cation-cation functions are superposable, andtheir minima and maxima are, respectively, located at the same position as the maximum andminimum of the cation-anion function There is therefore only one typical length, whichleads to a single peak (the principal peak) in the diffraction pattern of such salts

consider-As soon as multivalent cations are involved, we see in Figure 1.4that all the radialdistribution functions are strongly affected First, the first peak of the cation-anion radialdistribution function becomes much more intense and sharper This feature is often associ-ated with the formation of a well-organized first solvation shell For example, in LiF, Liþions can have between three and six Fin their first solvation shell, whereas in BeF2the

Be2þcan form tetrahedral BeF4arrangements only For these systems, we could show thatour simulations were in quantitative agreement with all the experimental data available in the

literature by reproducing the X-ray diffraction, and the infrared or Raman spectra of themelts (including the mixtures) [26–28] In many other melts, EXAFS is the only experimen-tal data available We then use a procedure [29] which allowed us to determine the speciation

of many multivalent cations such as La3þ in molten chlorides [30] or Zr4þ in moltenfluorides [31,32]

An advantage of molecular simulations is that we then have a detailed picture of the ture and of the dynamic processes which occur in the melt [27,33] For example, in LiF-BeF2

struc-melts, we could observe the simultaneous breaking/formation of ionic Be-F bonds by itoring the corresponding distances along the trajectory It is also possible to determine thelifetime of such bonds

mon-At longer range, the ordering of like-like ions is also largely modified compared to thesimple case of alkali halides The cation-cation and anion-anion radial distribution functionscannot be superposed anymore, as can be seen inFigure 1.4for BeF2and AlF3melts Theanion-anion function shows a first maximum at shorter distances, which still corresponds tothe first minimum of the cation-anion one Due to the higher repulsion between them, theminimal distance between two cations is larger Then, polarization effects start to play animportant role, as schematized inFigure 1.2 When the anion is more polarized, a largerbending of the M-X-M angle (M¼cation, X¼anion) is obtained, thus allowing the two cat-ions to approach more It was shown unambiguously for a series of divalent molten chloridesthat the smaller distances were obtained for small cations such as Zn2þ[34], as a conse-quence of the interplay between polarization and packing effects In these melts, a secondtypical distance of correlation arises; since it corresponds to a larger distance than thecation-anion one, it is generally called intermediate-range ordering Depending on theimportance of the effects and on the scattering length of the elements, it may give birth

to additional features in the diffraction patterns, which appear for smaller wavevector thanthe principal peak, and is generally referred to as a prepeak [35]

Figure 1.4 Radial distribution functions in three molten salts: LiF, BeF2, and AlF3.

In the case when the cation-anion bonds that are formed have a long-lived character, asshown inFigure 1.5for Be-F bonds, the system is said to form a network with long-rangeordering Characterizing such a network in experiments is not easy and computer simulationsare the ideal tool for understanding quantitatively the structure of the system Networks aremainly characterized by the coordination number of cations and by the nature of the linkagebetween two cations Depending on the number of anions that they share (i.e., which are inthe first coordination shell of both of them), we can observe some corner sharing (one con-nection), edge sharing (two connections) or face sharing (three connections) InFigure 1.6,

we show the progressive formation of a network of corner-sharing BeF4tetrahedra as theconcentration of BeF2is increased in LiF-BeF2mixtures

1.4 Dynamic Properties of Molten Salts

1.4.1 Viscosity

There are several routes for extracting transport properties from computer simulation iments They belong to two families, depending on whether equilibrium or nonequilibriumsimulations are employed Here, we will only detail the case of equilibrium simulations.Although those require longer simulation times than their nonequilibrium counterparts[36,37], they have a big advantage which is thatall the transport properties can be extractedfrom one simulation, provided that it is long enough [38,39]

exper-From an equilibrium simulation, the transport coefficients are usually obtained from theintegrals of time correlation functions through Green-Kubo relations The viscosity is deter-mined from [24]:

wheresabis one of the component of the stress tensor andab¼xy, xz, yz, xx-yy, or 2zz-xx-yy

In practice, the five resulting functions are calculated and averaged in order to ensure betterstatistics The viscosity is then given by the plateau value of the running integral An example

of comparison between our calculated values and the experimentally measured ones [40] is

Figure 1.5 Observation of a

fluoride ion exchange in the first

solvation shell of a beryllium ion

during a simulation.

shown inFigure 1.7for binary mixtures of LiF and BeF2 An excellent agreement is obtainedacross the whole composition range [27,33].

Depending on their structure, molten salts can have markedly different fluidities In thecase of mixtures either where no particular structural pattern is observed (molten alkalihalides) or which can be described as weakly connected structural entities (such as the ZrFx

species in molten fluorozirconates [31,32]), low viscosities are obtained (typically

1 mPa s1, i.e., of the same order of magnitude as liquid water at room temperature) As soon

as a strong network starts to form, the viscosity can become much higher In the case of

Figure 1.6 Progressive formation of a network in molten LiF-BeF 2 mixtures Top, Li3BeF5; middle, LiBeF3; and bottom, BeF2 On the left-hand side, the Be-F connections are shown with lines, while

on the right-hand side the BeF4tetrahedra are highlighted.

LiF-BeF2mixtures shown inFigure 1.7, this increase is very sharp By extracting the acteristic time from the stress tensor relaxation function and comparing it to the Be-F-Bebond forming/breaking relaxation time, we could show that the two quantities were stronglycorrelated, showing that this structural event was very likely to be at the origin of the vis-cosity increase [33] Interestingly, in the same study, we could show that the strong increase

char-of viscosity at large BeF2concentrations did not affect much the diffusion coefficient of therapidly diffusing Liþions, leading to an important decoupling of the dynamics of the system

to be taken due to the technique involved for dealing with long-ranged interactions The firstexpression was derived by Bernu and Vieillefosse in their study of the transport coefficients

of the one-component plasma [43], and we extended this work to the case of potentialsincluding polarization effects [44] In a first step, this approach was validated by comparingthe calculated values to experiments for a series of molten chlorides for which the measurewas possible These are reported inTable 1.1

Figure 1.7 Viscosity versus BeF 2

concentration in LiF-BeF2melts.

At 873 K, the simulations agree

very well with the available

experimental data [ 27 , 33 , 40 ].

In many molten salts, the experimental measure of thermal conductivity is difficult [46]

so that being able to predict it from computer simulations is of paramount importance.Contrary to the viscosity, it appears that this property is not well connected to any structuralfeature of the melts It was nevertheless possible to provide some general laws on its tem-perature and density dependence in a series of molten alkali halides [47]

1.4.3 Diffusion Coefficients

Although one can also use a Green-Kubo relation to determine the diffusion coefficients fromthe velocity autocorrelation function, it is often more convenient to use the Einstein relation,which links them to the long-time slope of the mean-squared displacements instead [1]:

is therefore possible to compare quantitatively our predicted values to a set of reference data.This has been done for LiF-KF mixtures, for which an excellent agreement was obtained [48],

as shown inFigure 1.8 Note that reproducing diffusion coefficients with such an accuracy,with a force field constructed without any empirical input, is not an easy task, for example,many force fields used for liquid water fail to reproduce this property at room temperature

In general, the diffusion coefficients of ions in molten salts are strongly related to theviscosity of the melt The more fluid it is, the higher they are Still we have observed that

in some network-forming liquids, decoupling effect can occur and it is then possible to havespecies with high diffusivities in a liquid of high viscosity [33] Similar effects have alsobeen observed in a series of alkali silicate melts [49], which share common features withthe LiF-BeF2melts discussed above

1.4.4 Electrical Conductivity

Finally, the electrical conductivity is a quantity that can vary substantially from one system

to another Although most molten salts are known to be good electrolytes thanks to theirhigh ionic conductivity, some of them such as pure BeF2 have very low conductivity

As for diffusion coefficients, we obtain the electrical conductivity from the long-time slope

of a plot of the mean-squared displacement of all the charges versus time,

s ¼ e2

kBTVtlim!1

16t

Table 1.1 Comparison of the Calculated and Experimental Thermal Conductivities for a

Series of Molten Chlorides

Unlike diffusion coefficients, for which the relation involves an average of the displacements

of all the ions of a given species, the electrical conductivity is a collective quantity and itscomputation involves all the displacements of all the species at a given time Its computation

is therefore more challenging, and longer simulations are required to calculate it We havealso recently set up some experiments for measuring electrical conductivities of molten fluo-rides, which provides us some comparison for our calculated data We generally get goodagreement [50,51], as shown inFigure 1.9for a LiF-NaF-ZrF4mixtures, but it appears fromour experience that for electrical conductivity, simulation results are more reliable sinceexperiments are difficult to perform

Although no systematic comparison of the conductivity of molten salts has been done up

to now,Figure 1.9shows the effect of adding a multivalent component to an alkali halidemixture: We observe a decrease of the electrical conductivity which is due to the formation

of ZrFxcomplexes As a result, the viscosity of the melt is increased, lowering all the fusion coefficients In addition, even if these complexes are relatively short lived [32], theirformation leads to a progressive reduction of the diffusion coefficient of Fwith the adding

dif-of ZrF4 The combination of these two effects results in a drastic reduction of the electricalconductivity of the melt

Figure 1.8 Diffusion coefficients of F and Liþions in LiF-KF mixtures at a temperature of 10 Kabove the melting point of each composition (sim, molecular dynamics; exp, PFG-NMR) [ 48 ].

1.5 Conclusion

Molecular modeling has now become a main tool for understanding the physical chemistry

of liquids Thanks to the development of advanced simulation techniques, this is particularlytrue in the case of molten salts, for which reliable predictions can be made entirely from firstprinciples

Of course molecular dynamics are intensively used in order to interpret experimentaldata Many improvements have been made in the instrumentation (EXAFS, NMR) inrecent years [52–54], and the application of these techniques to molten salts has provided

an unique source of data to which the simulations can be benchmarked Nevertheless, here

we have shown that more than a single picture, molecular dynamics provides a completemovie of the structure at short range (first solvation shell) as well as at longer ranges (for-mation of network), since it is possible to monitor the breaking/ formation of ionic bondsvery precisely; no experimental technique is yet able to provide such information for themoment

From the engineering point of view, it is very important to note that all the importanttransport quantities (diffusion coefficients, electrical conductivities, viscosities, and eventhermal conductivities) can be obtained provided that the simulations are long enough.The case of static properties was not detailed here because much more experimental dataare available in the literature; nevertheless, it is worth noting that our simulations resultscan also be used as an input for thermodynamic modeling [55] If efficient bottom-upstrategies are set up, in the future, it will be possible to extract the thermodynamics(including phase diagrams [56,57]) of any system under any temperature or compositionconditions very straightforwardly ab initio, i.e., without involving any experimentalinformation

As for the perspectives, we believe that important developments will appear for themodeling of electrochemical properties From the practical point of view, most molten saltsapplication concern electrochemistry [18–21,58–60] and it is therefore crucial to be able tohelp interpreting such experiments First steps have been made toward this direction since it

is now possible to calculate activity coefficients of lanthanide and actinide species in moltensalts [61] Simulations of the interface between a molten salt and a metal electrode have alsobeen performed [62,63] In the future, being able to model electrochemical reactions will benecessary for guiding the development of future devices involving molten salts

Figure 1.9 Calculated (filled symbols) and measured (open symbols) conductivities for a LiF- NaF mixture with an added amount

of ZrF4[ 50 ].

[7] Frenkel, D (2012) Simulations: the dark side arXiv:1211.4440.

Picard, G (2009) Reactor physics and reprocessing scheme for innovative molten salt reactor

University Press (Cambridge).

[24] Hansen, J.-P., McDonald, I.R (2006) Theory of simple liquids Academic Press (Amsterdam).

11461–11467.

Okamoto, Y., Vivet, F., Bessada, C (2010) In situ experimental evidence for a

114, 6472–6479.

204511.

[39] Dewan, L.C., Simon, C., Madden, P.A., Hobbs L.W., Salanne, M (2013) Molecular dynamics simulation of the thermodynamic and transport properties of the molten salt fast reactor fuel LiF-ThF4 J Nucl Mater., 434, 322–327.

104507.

diffusivity of molten alkali halides by the forced Rayleigh scattering method 1 Molten LiCl,

[47] Ohtori, N., Oono, T., Takase, K (2009) Thermal conductivity of molten alkali halides:

44–48.

1500 K: a powerful tool to investigate the dynamics and the local structure of inorganic melts.

extraction processes in molten fluoride media: application to nuclear spent fuel reprocessing.

1177–1183.

2 Raman Spectroscopy and Pulsed

Neutron Diffraction of Molten Salt Mixtures Containing Rare-Earth

Trichlorides: Trial Approaches from Fundamentals to Pyrochemical

Reprocessing

Yasuhiko Iwadate

Graduate School of Engineering, Chiba University, Chiba, Japan

2.1 Introduction

It is desirable to restrain combustion of fossil fuels and make safe use of nuclear energy

as global warming advances Continuous and persistent efforts have been made frompolitical and technological standpoints to develop novel processing of spent nuclearfuels for nuclear nonproliferation and peaceful uses of nuclear energy As pointed outelsewhere [1], high-temperature molten salt technology has so far been applied to irra-diated nuclear fuel processing There are currently two developed pyrochemicalprocesses operating at the Research Institute for Atomic Reactors (RIAR) in Russiaand at the Argonne National Laboratory (ANL) in the United States, respectively Thetwo facilities are both able to process irradiated nuclear fuel electrochemically, usingmolten chloride melts such as NaCl-KCl eutectic or NaCl-2CsCl at RIAR and LiCl-KCl eutectic at ANL Advanced molten salt technologies are currently being developed

at both the ANL [2–4], the RIAR [5], and at several research institutes in Japan andacross Europe [6–10]

Much attention has recently been focused on pyrochemical reprocessing of oxide fuels

in molten salts with special composition and combination to make UO2 þand PuO

2 þions

coexist stably However, several scientific problems still remain unsolved in utilizing thistechnique [11,12], for instance, elements of fission products coexisting in molten salts hinderthe sustenance of electrolysis or codeposition It is thus important that structural analyses

of molten salts are carried out to comprehend codeposition phenomena and to clarify themechanism blocking the codeposition of UO2and PuO2

It is well known that numerous kinds of rare-earth elements are produced in the fissionprocess of uranium, which are often utilized as substitute elements in fundamental studiesand industries related to nuclear energy As a representative material for electrolysis, LaCl3

and YCl3were selected for the sake of budgetary cutback and safety

Molten Salts Chemistry

© 2013 Elsevier Inc All rights reserved.

The goals of the present study, which is preliminary, to put this processing intoeffect, are to analyze the local structure of NaCl-based mixed alkali chloride melts assolvents and LnCl3-added melts as solutions instead of UO2 þand/or PuO

2 þ-containing

melts according to Raman spectroscopy and pulsed neutron diffraction (ND) and to selectthe most appropriate solvent system, as well as to make an accurate forecast of thereprocessing

2.2 Experimental

The chemical LnCl3(Ln¼L and Y) was synthesized from mixed powders of Ln2O3and NH4Claccording to the following reaction at 650 K: Ln2O3þ6NH4Cl!2LnCl3þ6NH3þ3H2O.After chlorination, continuous evaporation of crude LnCl3 was allowed at 1300 K(or 1273 K) for 8 h under reduced pressure The purified material, without impurities such asoxide and oxychlorides, was deposited on the water-cooled quartz surface [13] The purifiedLnCl3sample was sealed in a transparent cylindrical fused silica cell of 4 mm inner diameter,

6 mm outer diameter, and 130 mm height to prevent spoilage since LnCl3is primarily ratherhygroscopic and reacts with water vapor to form LnOCl at elevated temperature The reagentgrade alkali chlorides were desiccated in the usual manner, melted under dry N2atmosphereand solidified The mixtures of CsCl and NaCl in adequate quantities were prepared byaccurately weighing the components in a glove box filled with dry N2gas, followed by drying

at 473 K for 2 h, melting, and rapidly quenching so as to obtain good homogeneity The mixtureswere introduced into the transparent cylindrical fused silica cells mentioned earlier in the glovebox and sealed under reduced pressure

Raman spectra were measured with a specially redesigned laser Raman eter (original apparatus: NR-1800, JASCO, Japan) equipped with a triple monochromatorusing an Arþion laser at the wavelength of 514.5 nm with an output of 200 mW as an exci-tation light source All the samples were measured three times to obtain the spectra of simplyscattered light, of lights scattered parallel (IVV) and perpendicular (IHV) to the polarization ofthe incident light The obtained spectra were then smoothed, normalized, and reduced in themanner previously reported [14] The details of Raman scattering experiments and data ana-lyses have previously been described [15,16]

spectrophotom-The purified and granular salt samples with well-defined densities [17] were introducedinto the transparent fused silica cells of 8 mm inner diameter, 8.6 mm outer diameter, and

65 mm height (beam irradiated part) in the glove box and sealed under reduced pressure,and pulsed neutron scattering experiments were performed at different temperatures forevery binary and ternary melt using the High Intensity Total scattering spectrometer, which

is designed to measure the structure factorS(Q) at a high rate of data collection over a widerange of momentum transferQ (from 5 to 500 nm1in this work) and installed in the pulsedneutron source of High Energy Accelerator Research Organization, Neutron Science Labo-ratory at Tsukuba, Japan The neutron scattering intensities were measured with sevencounter banks at angles of 7, 13, 20, 30, 50, 90, and 150by the time-of-flight method.TheS(Q) was obtained from the measured intensity after some corrections such as sub-traction of cell intensity, absorption, multiple scattering, normalization with a standardvanadium rod and so on These correction procedures have already been reported in detailelsewhere [18,19] The coherent scattering lengths of component elements with naturalabundance were taken from the literature [20]

2.3 Results and Discussion

2.3.1 Raman Spectroscopy

Prior to the consideration of experimental results, theoretical background and procedure fordata acquisition and analysis are described briefly as follows All the measured spectra weresmoothed in the manner previously reported [14] and normalized by the equation:

Inormal¼ I ofð Þ  Iming= Ifmax Iming (2.1)whereI(o) is the measured Raman intensity at the Raman shift o, Iminthe minimum intensity

of theI(o), and Imaxthe intensity of the measured Raman spectra at 20 cm1in this work,and thus the intensities of the spectra can be compared with one another for the differentcompositions As the contribution of Rayleigh scattering was included even over the highwavenumber region in each spectrum, the reduced Raman intensityR(o) was extracted fromthe measured Raman intensityI(o) in terms of the equation [15,16],

Rð Þ ¼ I oo ð Þo oð 0 oÞ4½nð Þ þ 1o 1 (2.2)whereo is the Raman shift mentioned before, o0the excitation laser frequency, andn(o)þ1the Boltzmann thermal population factor defined asn(o)þ1¼[exp(hoc/kT)1]1þ1 with

h being the Plank constant, c the velocity of light, and k the Boltzmann constant, respectively.Only a Raman band for a symmetric vibrational mode has a depolarization ratior much lessthan 0.75 and thus contributes to the isotropic intensity Iiso as a consequence The IVV

spectrum contains information on both the isotropic and anisotropic polarizability tensorcomponents; on the other hand,IHVspectrum contains only contributions from the aniso-tropic polarizability tensor components The spectrumIisodue to the isotropic polarizabilityelements can be obtained by subtraction [15,16],

Iisoð Þ ¼ Io VVð Þ o 4

The anisotropic spectrumIanisois substantially equal toIHV

Light scattering experiments were performed at first to study solvent properties of binaryalkali chloride melts with NaCl As exemplified by NaCl and CsCl, alkali chlorides arestrongly ionic in the type of chemical bonding In the unit cube of crystalline NaCl arrange-ment, each Naþhas six equidistant Cls as nearest neighbors, and vice versa, called anoctahedral coordination [21] In crystalline CsCl grouping, Csþhas eight Cls in the nearneighborhood [21] The coordination numbers of Claround Naþand Csþin the corre-sponding alkali chloride melts have been reported to be 3.7 and 4.6, respectively [22] Drasticchanges in local structure such as coordination numbers and interatomic distances surelyoccurred on melting of each salt [22], being therefore of much interest to investigate theshort-range structure of mixed alkali chloride melts according to Raman spectroscopy Asfor 2CsCl-NaCl melt, for instance, no typical sign of Raman scattering was observed inthe spectra up to 450 cm1, but a simple and unique wing of Rayleigh scattering decreasedmonotonously in intensity with increasing wavenumber Only the existence of Rayleighwing indicated that there were no ionic association in the melts and no formation of novelspecies, neither ionic nor molecular The similar spectra were observed for other mixture

melts Consequently, simple mixing of Csþ, Naþ, and Clions was thought to appear in themixture melts, irrespective of the differences in ionic size and concentration.

LaCl3 crystal is hexagonal, in which ions have been placed in the positions of C6 2

(P63/m) [21], that is, each La3þis surrounded by nine Cls, where six of these La3þ-Cldistances are short and the other three are long An X-ray diffraction study of molten pureLaCl3has reported that an La3þion is surrounded by six Clions to form complex ions such

as LaCl6 , La

2Cl11 5, and so on [22,23], as is the case with molten CeCl

3[24] However,another structural image has been proposed in the works by ND [25,26] and moleculardynamics (MD) [27,28], in which the nearest neighbor coordination number is concluded

to change with cationic size in the pure rare-earth trichloride melts

In the ternary (2CsCl-NaCl)-LaCl3system, apparent changes were observed with ing LaCl3 in the intensities of scattered lights as shown inFigure 2.1 Broad bands wereobserved at about 150 to 320 cm1and shoulders around 70 to 170 cm1 In order to assignthis band some normal mode of vibration, a VV Raman spectrum and an HV one weremeasured, in which the polarization planes of an incident light and a scattered one areparallel to each other, and those are shifted by 90, respectively The VV and HV spectraare depicted inFigure 2.2, where the broad bands at 150-320 cm1in the VV spectra areattenuated largely in the HV ones, but those at 70-170 cm1are not necessarily weakened.Using both VV and HV spectra, isotropic and anisotropic components of the Raman bandswere estimated, as can be seen fromFigure 2.3 The maxima in the isotropic componentswere thought to be due to the totally symmetric stretching vibrationn1of octahedral speciesand/or its associated species This result supports the sixfold coordination scheme [29] asproposed in the pioneering work of Papatheodorou [30,31]

increas-If each La3þis surrounded by six Cls in ternary melts to form an octahedral-type ture, there must be two more Raman-active vibrational modes, that is, one is the degeneratestretching vibration mode n2 and the other the degenerate bending vibration mode n5.The lack of those vibrational modes has so far been thought to be due to the weakness inintensity and the instability of the species at elevated temperature This is partly becausethe octahedral structure derived from X-ray or ND is time- and space-averaged and there

struc-is a wide dstruc-istribution from 3 to 9 in the real coordination number [32] As illustrated in

Figure 2.1 Variation of Raman spectra with LaCl 3

concentration in molten (2CsCl-NaCl)-LaCl3system at 1073 K

(LaCl3at 1173 K).

Figure 2.4, it was found from precisely deconvoluting the anisotropic components of Ramanbands such that each anisotropic component could be represented by superimposing twoprofiles with peaks; one is an2band centered at 225-250 cm1and the other an5band at100-125 cm1 It was thus clarified that there existed stable octahedral-type geometricarrangement of ions, that is, complex ions and their polymeric ions in the mixture melts.Similar tendencies were observed in other alkali chloride solvent systems Finally, the effect

of alkali cation on the stability of octahedral associated species is discussed As depicted inFigure 2.5, the stability of octahedra increased with increasing radii of alkali cations, which

Figure 2.2 Variation of VV and HV spectra for molten (2CsCl-NaCl)-LaCl3system at 1073 K (LaCl3at 1173 K).

Figure 2.3 Variation of isotropic and anisotropic components

of reduced Raman spectra with LaCl3concentration in molten (2CsCl-NaCl)-LaCl3system at 1073 K Solid line, isotropic; dashed line, anisotropic.